mysql: SQL grouping rows in a source database table, generate only grouped rows by adding a group row

I am trying to write a SQL query to work on a raw feed that has been imported into a MySQL database table. The rows in the table look slightly like this example.

| sku | group_name | description | size | type |
| ------------ | ------------ | ------------------- | --- --- | -------- |
| EX-1000-XS | NULL | Long text | XS | single
| EX-1000-S | NULL | Long text | S | single
| EX-1000-M | NULL | Long text | M | single
| EX-1000-L | NULL | Long text | L | single
| EX-1001-M | EX-1001 | Another long text | M | single
| EX-1001-L | EX-1001 | Another long text | L | single
| EX-1001 | NULL | | | group |

I'm trying to get those that do not have a group name, but with the name of the group set, along with a group. Ignoring things that were already in a group.

| sku | group_name | description | size | type |
| ------------ | ------------ | ------------------- | --- --- | -------- |
| EX-1000-XS | EX-1000 | Long text | XS | single
| EX-1000-S | EX-1000 | Long text | S | single
| EX-1000-M | EX-1000 | Long text | M | single
| EX-1000-L | EX-1000 | Long text | L | single
| EX-1000 | NULL | | | group |

I know I can get a group row using GROUP BY, and then I UNION? How could you refer to the group to replace that? NULL, I'm trying REGEXP_SUBSTR in the column sku.

I set up a violin with the example, https://www.db-fiddle.com/f/ua6iwXGHcPDVKbCkqHvs2E/0